After years of working inside some of the world’s largest media agencies, Senan Mele joined BIA Advisory Services as VP of Forecasting and Data Analysis in April.
Mele began his media career in analytics, quickly evolving from basic reporting to mastering tools like R, Python, and SQL to automate workflows and support strategic forecasting. Over the years, he built and managed complex data systems at agencies including MEC, Carat, m/Six, McCann, and Horizon. As clients demanded deeper insights and more efficient media spend, Mele shifted his focus to predictive modeling and budget optimization – skills he now brings to his role at BIA.
In this conversation with Radio Ink, Mele shares his vision for modern forecasting, why data quality is the bedrock of successful AI integration, and how radio remains a crucial touchpoint in a fragmented media world.
Radio Ink: You have plenty of buy-side experience – how does this all culminate in your new role with BIA?
Senan Mele: So BIA has been forecasting for many, many years, but the goal here is to expand on the strong foundation that’s already there.
My goal specifically is to bring more data sources into the forecasting models that BIA has, looking at a lot more economic data points and seeing how changes in those data points impact ad revenues. And helping further streamline existing systems and create more automation than there already is across the team.
So that’s really the impact that I’m planning to have at BIA – to help BIA’s clients anticipate change, come up with sensitivity analysis and scenario planning: if this economic indicator changes in one way or another, this is what we can expect in changes in ad revenue.
One of my early priorities at BIA will be to deliver a sensitivity analysis across various economic scenarios to reflect concerns around tariffs and their effect on local advertising. Leaders are making decisions on tighter budgets and shorter lead times. Scenario-planning helps them stay ready for what’s next, which is especially important when it comes to understanding how tariffs and other economic shifts might impact local ad revenue.
Radio Ink: You’ve worked with global brands on campaigns across multiple continents. How does that help shape your understanding of media behaviors and audience trends locally with BIA?
Senan Mele: I’m in my fourth week now and definitely still in the process of acclimating to the ways the team does forecasting, but what I’ve noticed over time is that media spend has been fragmenting over a lot of different channels. It’s happened at the global level, and we’re seeing it in the local ad space as well.
Before, there was a lot of focus on what we call legacy media. But there are a lot of new technologies that have been coming out, like streaming and digital ad platforms, short-form social content like TikTok, YouTube Shorts, Roku, Amazon Fire TV, Samsung Ads, and programmatic. And let’s not forget AI and how AI is being used in optimizing campaigns now.
But the point I’m trying to make is I’m seeing that there is a fragmentation of ad revenue, and we need to be able to account for that in forecasting. BIA is already doing that. My goal is to dive deeper into it at a more granular level and be able to account for those changes without forgetting legacy media. Even though all of these new channels and media types have been eating into legacy media’s market share, legacy media like radio are still a big part of ad spend and need to be accounted for.
But something that’s interesting will be to see – and to predict, as much as we can – how that change will continue evolving and the impact that AI will have on all of these media channels, including traditional radio and TV, as people are starting to use things like large language models to make decisions.
Radio Ink: On the sales side in radio, we’re seeing more teams leaning into selling digital, particularly Over-the-Top and Connected TV, and then drawing people into AM/FM advertising. How do you see radio’s role evolving in that omnichannel media strategy?
Senan Mele: Since TV came out, radio has always faced predictions that it was a dead medium. As digital channels came out, there were predictions that TV and radio were going to be gone. But I think radio and TV will continue to play a fundamental role in consumer behavior.
I don’t think it’s wise for brands to just say, “Oh, we don’t need radio” or “We don’t need TV” and only focus on digital platforms. Of course, the younger generation is on digital platforms, sometimes not on traditional platforms. But there’s also a convergence of both digital and traditional platforms that’s happening, because brands need to reach people not in the digital world. They’re often on the go or doing different things, and they’re not on the phone or computer. They’re listening to radio while running errands.
Brands need to be able to share that message and reach those people where they are as they go through their day-to-day. And so I think radio will continue to play a significant role in the media landscape, but we’ve already reached the point where it’s important for brands to reach people across different channels.
Radio Ink: At m/Six you worked directly with platforms like Facebook, Google, and even Twitch. What did you learn from those tech relationships, and how has that shaped your overall media worldview?
Senan Mele: In the digital landscape, a lot of platforms are often claiming credit for conversions. So let’s say we have Facebook and Google Search. What happens is, as marketers are tracking conversion, there’s this issue of allocating conversion to specific media channels.
Facebook might say, “Hey, I contributed 80 sales,” and Google Search might say, “Actually, I contributed 50 sales,” and a third platform might claim 30. But let’s say the total number of sales for the month was actually 100. When we add up what all the platforms are claiming, it ends up being more than 100 sales.
So something I’ve learned – and this is a big issue in the media industry, especially on the buy side – is that all of these platforms are claiming the same sales events. So, how do we make sure we are allocating sales correctly across each of these platforms and that one platform isn’t claiming sales it didn’t truly generate?
These platforms aren’t being intentionally dishonest; it’s also tied to things like retargeting. For example, someone sees an ad on Facebook, then later they’re searching on Google, and they buy the product. Facebook says, “Well, they saw it on our platform first,” while Google says, “They bought it after they searched with us.”
So both platforms played a role. The person saw the ad on Google and immediately bought it, but Facebook helped build awareness.
Different strategies have been used to try and solve this. There’s first-touch attribution, where credit goes to the platform that made the first impression. There’s last-touch attribution, where it goes to the last one before the sale. Some models spread the credit evenly. Others rely on statistical modeling.
That’s where I really started to see how confusing the landscape can be, especially with how fast everything moves in digital. Moving away from traditional media made it murkier to understand which platform truly contributed to a sale because consumers interact with all of them at once – and that’s where statistical modeling and machine learning come in.
It’s an anticipatory approach, not just a reactionary one.
Radio Ink: What’s a question you think that broadcasters and media buyers aren’t asking yet, but maybe they should be when it comes to the future?
Senan Mele: Many companies are asking, “How do we use AI?” but they should be asking, “How do we get our data into a strong, usable state so it can work effectively with AI?”
It’s not just about adopting AI for the sake of innovation. There’s a risk of creating a kind of detachment if we treat AI as a quick fix or a box to check. The real goal should be more fundamental: before we even think about implementing AI, we need to go back to data. A lot of companies are racing to integrate large language models, but I don’t think enough of them are going back to the basics to evaluate whether their data is clean, accurate, and formatted in a way that actually supports those models.
With the rise of AI, the importance of well-cleaned and accurate data is having a resurgence. Because we have this new technology that’s out, but it’s extremely dependent on the basics. As the saying goes, “wrong data in, wrong data out.” So I think that’s a question companies should be asking themselves, and broadcasters, everyone who is interested in the new technologies should first go back to the drawing board and make sure they have good, accurate, strong data and data systems in place.
This interview has been edited for length and clarity.
